On Creating Depth Maps from Monoscopic Video using Structure from Motion